Agronomía Costarricense 28(2): 37-51. 2004

PROGENY TEST ANALYSIS AND POPULATION DIFFERENTIATION OF MESOAMERICAN MAHOGANY ( macrophylla)

Carlos Navarro1/*, Gustavo Hernández*

Palabras clave: Swietenia macrophylla, heredabilidad, fitomejoramiento, evaluación de progenies, poblaciones naturales. Keywords: Swietenia macrophylla, heritability, tree improvement, progeny test, natural populations.

Recibido: 07/01/04 Aceptado: 29/09/04

ABSTRACT RESUMEN

The performance of open-pollinated Análisis de experimentos de progenies single tree families of mahogany Swietenia (familias) de una colección Mesoamericana de macrophylla from populations in Mesoamerica caoba (Swietenia macrophylla). El desarrollo de was evaluated in 3 trials established in northern progenies de árboles madre de polinización abier- . The trials at and Lagartera ta de caoba, Swietenia macrophylla, fue evaluado () contain families of Costa Rican en 3 ensayos establecidos en la Zona Norte de origin, while the Laberinto (Los Chiles) trial Costa Rica. Los de Upala y Lagartera (Los Chi- contains material from 6 Central American les) incluyen familias originarias de Costa Rica, countries and Mexico. Data on root-collar mientras que el de Laberinto (Los Chiles) presen- diameter, total height, survival, and ta material de 6 países centroamericanos y Méxi- grandella attack were collected. The analysis co. Se recolectó información sobre diámetro a la indicate significant family and population base, altura total, sobrevivencia, y ataque de differences for height and diameter, but H. Hypsipyla grandella. Los análisis indican dife- grandella attacks were uniform over all sites. rencias significativas a nivel de progenie y proce- Heritabilities at 1.7 years for Upala were dencias para diámetro y altura, pero el ataque de 0.54±0.02, and 0.55±0.02 for diameter and H. grandella fue uniforme en todos los sitios. Las height, respectively; after this measurement this heredabilidades a los 1,7 años para Upala fueron trial was burned completely as a result of drought de 0,54±0,02 y 0,55±0,02 para diámetro y altura in El Niño year, so further measurements could respectivamente. Después de esta medición el en- not be made. Lagartera at 0.7 years presented sayo se quemó por completo, debido a sequías ba- heritabilities for diameter and height of jo el efecto de El Niño, razón por la cual no pu- 0.55±0.008 and 0.57±0.008. Laberinto presented dieron efectuarse mediciones posteriores. El heritabilities of 0.48±0.01 for diameter (2.9 ensayo en Lagartera, de 0,7 años, presentó here- years), 0.6±0.01 for height, 0.1±0.002 for H. dabilidades para diámetro y altura de 0,55±0,008 grandella attack; 0.07±0.002 for number of y 0,57±0,008. En Laberinto, las heredabilidades shoots, and 0.18±0.003 for stem form 2.7 years obtenidas fueron de 0,48±0,01 para diámetro (2,9

1/ Autor para correspondencia. Correo electrónico: * Tropical Agricultural Research and Higher Educa- [email protected] tion Center (CATIE), Turrialba, Costa Rica. 38 AGRONOMÍA COSTARRICENSE after planting. Flooding in 1998 damaged the años), 0,6 ± 0,01 para altura, 0,1±0,002 para el Lagartera trial, which was also severely attacked ataque de H. grandella; 0,07±0,002 para número by H. grandella. This resulted in very low de ejes, y 0,18±0,003 para la forma a los 2,7 años heritabilities, with large standard errors; de plantado. Las inundaciones en 1998 dañaron therefore its genetic values in the first el ensayo de Lagartera, que además fue severa- measurement are considered unreliable. The mente atacado por H. grandella. Esto se manifes- plantation recovered, and after 3 years genetic tó en heredabilidades muy bajas con errores es- values were comparable with the other 2 trials. tándar altos, por lo tanto los valores genéticos en Isolated mother trees produced slow-growing la primera medición se consideran poco confia- families in most cases, in comparison with the bles. La plantación se recuperó y después de 3 clustered ones or those in natural dense forests, años los valores genéticos fueron comparables suggesting inbreeding mechanisms. con los otros 2 ensayos. En general, los árboles madre que estaban solitarios, produjeron proge- nies de crecimiento lento, en comparación con las progenies de árboles madre que estaban en grupos o en bosque natural, lo que sugiere meca- nismos de endogamia.

INTRODUCTION (Matamoros and Seal 1996), while current logging of natural populations for the Swietenia macrophylla is an important international market occurs in Brazil and Bolivia. tree species in the neotropics, it is found in the These tree species present a low rainforests between latitudes of 22o North and proportion of adult trees, which joined with a low 20o South of the Equator. Graham (1999) natural regeneration rate, increase their rareness. indicates that pollen from an ancestor of the The extensive harvest of S. macrophylla Meliaceae family was present in the Pliocene and for its valuable wood has resulted in high Miocene in Mexico, which led us to believe that concern over its conservation status and the species was present there many years ago. sustainable use (Proposed Amendment to CITES In Meliaceae, Swietenia is the most Appendix II 1997), and a strong focus of current important genus for wood production followed by research on these topics (Negreros and Mize . Beginning in the 19th century and until 1994, Gullison et al. 1996, Snook 2003, Navarro now, the mahoganies have been the pillar of the et al. 2003). The FAO is establishing a network forest industry of Meso and South America. They to facilitate the genetic conservation of S. cover the neotropical territory from Mexico to macrophylla, and other species of the Meliaceae Brazil and Argentina, and the Caribbean Islands. family (Patiño 1997). S. mahagoni has already During the last decades, the Swietenia been heavily exploited, and both S. mahagoni natural populations have been severely affected and S. humilis were listed under Appendix II of and reduced by several factors, mainly due to CITES in 1992 (Patiño 1997). deforestation processes that diminish populations In 1994, CATIE and ITE (Institute of as well as selective logging that affects the Terrestrial Ecology) together with other genetic makeup of populations. institutions and the European Union support, The Mexican and Central American made a collection of mahogany germplasm in populations have been heavily exploited seven Mesoamerican countries. With this NAVARRO Y HERNÁNDEZ: Population differentiation of Mahogany 39 material studies of genetic variation were done, orientation and for thorough documentation of gene banks and progeny trials in Costa Rica and the collecting mission. Mexico were established. This work have been Collections were made from 42 different done considering the importance of using both mahogany populations, ranging from Mexico to molecular and quantitative markers for gene Panama. The number of trees sampled within conservation and breeding. populations varied according to its size and This paper investigates three major aspects: accessibility. Populations in each country were (1) The genetic variation throughout Mesoamerican located using the expertise of local collaborators populations and families of Swietenia macrophylla; and previous reconnaissance. The approximate (2) The growth variation in families and populations extent of each population was gauged as of mahogany; (3) The influence of collecting objectively as possible. isolated trees in their progeny performance. Trees were sampled along a transect, the The shootborer Hypsipyla grandella is a initial bearing was randomly selected. Mahogany Lepidopteran that co-evolved with some Meliaceae trees were either solitary or clumped. When trees species; it attacks preferentially; Swietenia, were clumped, collections were restricted to Cedrela and Carapa. Therefore, the adaptation of individual trees more than 100 m apart. To obtain different Swietenia populations to the shootborer maximum diversity and avoid seeds from related attack was also studied in these experiments. trees, only five individual trees per clump were collected. Up to a maximum of 50 trees were collected within each population along the MATERIALS AND METHODS transect. Only mature trees were selected. In some cases, populations were so sparse that only Collection solitary trees could be collected. Herbarium material was dried and Field collection of leaves, herbarium mounted at CATIE’s herbarium, and seeds were material and seeds from natural Swietenia dried and stored at the CATIE seed bank. populations were carried on. The populations Table 1 and figure 1 show the sites of sampled covered a wide range of environments, collections in Central America and Mexico. population densities and degrees of exploitation. This is the most extensive single-tree Prior to collection, general information collection of mahogany that has been made in was sought on: (1) climatic data - including this area. The Institute of Tropical Forestry in topography, geology, soil, vegetation, land use; Puerto Rico made an earlier provenance and (2) socio-economic data - including collection of S. macrophylla in 1964 and 1965. population, agricultural surveys, economic At that time, 14 provenances were collected in indicators, and information on the infrastructure, Mexico and Central America (Boone and roads and other means of transportation. Chudnoff 1970). This information was used to: Field trials ● Define eco-geographically distinct areas in the distribution of the species for To observe variation in the quantitative sampling; genetic parameters, 3 of the 6 measurements ● Estimate the likely extent of within-species (before and after the strong attack of Hypsipyla variation, based on the heterogeneity in its grandella) are presented in this article. Details natural distribution area; about experimental design and dates of ● Assess the threat of genetic erosion; and measurement are shown in table 2. ● Predict the best timing for collecting. In the first measurement, root collar diameter, total height, and H. grandella attack Information concerning the human and were analyzed, while variables of stem form physical environment was also useful for field and attack response were added in the last one. 40 AGRONOMÍA COSTARRICENSE

Table 1. Populations of Swietenia macrophylla sampled in Central America and Mexico. Latitude and longitude are given in decimal degrees.

Country Population Latitude (N) Longitude (W) Altitude (masl) No. trees sampled

Mexico Naranjal, Quintana Roo 19.36676 88.46131 50 15 Nuevo Becal, Campeche 18.80756 89.32721 150 46 San Felipe, Quintana Roo 18.74599 88.35480 50 20 Escarcega, Campeche 18.60068 90.82924 50 9 Laguna Kana, Quintana Roo 19.44662 88.44116 50 5 Madrazo, Quintana Roo 18.03352 89.24794 150 5 Belize Las Cuevas, Cayo 16.75180 89.00106 600 10 San Pastor, Cayo 16.70621 88.97249 600 26 New María, Cayo 16.82337 89.00043 600 13 Grano de Oro, Cayo 16.71945 89.01754 600 25 Río Bravo, Orange Walk 17.84534 89.03319 50 35 Guatemala Bethel, Petén 16.48350 90.50282 120 32 Tikal, Petén 17.22520 89.61216 250 56 La Técnica, Petén 16.91341 90.91274 125 47 Bio-Itza, Petén 16.85234 90.93113 20 2 Honduras Corrales, Colón 15.51259 85.94703 650 12 Lancetilla, Atlántida 15.73991 85.45721 30 35 Mangas, Colón 15.51259 85.94703 680 1 Comayagua, Comayagua 14.46010 87.68314 500 5 Otoro, Siguatepeque 14.52160 88.00089 600 3 Terciopelo, Sahsa 14.00675 83.93559 60 26 Mukuwas, Bonanza 14.04645 84.49976 200 38 Costa Rica Marabamba, Los Chiles 10.94656 84.63752 45 67 Caño Negro, Los Chiles 10.91774 84.42994 55 37 Santa Cecilia, La Cruz 11.06101 85.27857 300 12 Upala, 10.53508 85.08336 50 13 , Liberia 10.53364 85.35957 270 37 Playuelas, Los Chiles 10.92175 84.69872 35 4 San Emilio, Los Chiles 10.97147 84.77328 30 64 Abangares, Guanacaste 10.06600 84.49641 50 6 , Alajuela 9.55140 84.29620 250 1 Turrubares, Alajuela 9.51212 84.31335 350 1 Chapernal, Puntarenas 10.07469 84.82586 50 10 Panama Quintín, Darien 8.22469 78.08581 70 10 Punta Alegre, Darien 8.26119 78.23616 10 5 Tonosí, Los Santos 7.44882 80.29070 100 15 Gatún, Gatún 9.26800 79.91958 20 4 Paraíso, Paraíso 9.03278 79.62656 50 1 Balboa, Ancon 8.95500 79.95430 50 1 Summit, Ancon 9.06489 79.64622 50 3 Calabacito, Veraguas 8.24636 81.08095 50 1 Coiba, Veraguas 7.50102 81.69603 10 1 Cerro Hoya 7.32321 80.59560 500 23

Total 782 NAVARRO Y HERNÁNDEZ: Population differentiation of Mahogany 41

Fig. 1. Mesoamerican distribution of the populations planted in the trials of Swietenia macrophylla.

Table 2. Field trials data.

Trial Location Experimental design Planting date Date of Spacing (m) (GPS) measurements in days after planting

Upala 10.53409 N, 15 blocks, 31 families, 2 trees 12th June 1996 621 3 X 3 85.02287 W per plot, in lines with Lagartera, 10.58853 N, 15 blocks with 31 families and 19th July 1996 585, 1658, 1812* 3 X 3 Los Chiles 84.43002 W 3 trees per plot. Laberinto, 10.94774 N, 15 blocks with 91 families and 14th June 1997 251, 979, 1054*, 1327 3 X 3 Los Chiles 84.70994 W 2 trees per plot.

* only for diameter

The experiments were measured for total height Laberinto, established 1 year later, contains (cm) from the ground to the terminal shoot, populations from Central America and Mexico diameter was measured at the root collar and the (Table 4). Laberinto has some populations in attack of H. grandella was taken as a discrete common with trials at Lagartera and Upala. variable indicating presence or absence of attack. For the measurements taken during 2000 The trials were established at three sites in and 2001, only the trials at Laberinto and northern Costa Rica. Trials at Lagartera and Upala Lagartera were evaluated, and the new variables (Table 3) contain Costa Rican families. The trial at were evaluated as: 42 AGRONOMÍA COSTARRICENSE

Table 3. Costa Rican families planted in trials at Lagartera (Los Chiles) and Upala.

Population Nº mother trees (Family) Latitude (GPS) Longitude (GPS)

Caño Negro I 643, 644,645, 646, 647, 648, 649, 650 10.94774 84.70994 Caño Negro II 651 10.94705 84.72119 Caño Negro III 652 10.94646 84.72319 Caño Negro IV 653 10.94774 84.70994 Playuelas I 658 10.91815 84.69996 Playuelas II 660 10.92770 84.68567 Playuelas III 661 10.92635 84.68893 Marabamba I 665* 10.95426 84.65771 Marabamba II 668, 673, 674,6 76 10.95426 84.65771 Marabamba III 680, 681 682 10.95403 84.65751 Marabamba IV 683 10.97844 84.70361 San Emilio I 684, 685, 686, 687,688, 689 10.97235 84.77287 San Emilio II 690 10.95402 84.77411 San Emilio III 691 10.53615 84.47122 San Emilio IV 692 10.58191 84.46671 San Emilio V 693 10.95971 84.77346

* Family 65 was planted only at Lagartera. Roman population numbers indicate different grouping of trees within the collection transect.

● Form: 1 = straight 2 = leaning 3 = crooked Moist Forest with climatic association. This life 4 = coppice 5 = stump zone is characterized by having a biotemperature between 24 and 25°C and around 2500 mm of ● Response to the attack: annual rainfall with a dry period of 1 = One shoot after the attack approximately 3 months. 2 = Two shoots (bifurcation) after the attack The Laberinto trial was established in 3 = Three shoots after the attack northern of Costa Rica in soils classified by 4 = Four or more shoots after the attack Perez et al. (1978), as belonging to the Typic 5 = Bad stem form by other damages Tropaquept family fine silty isohyperther- mic. These gley and humic gley soils are For both measurements the variable found in drained alluvial valleys, and show isolation (solitary) of the mother tree was added, little evolution and presence of organic considering : 1. Isolated mother tree (no other accumulations on the surface. The trees of the same species) at a distance of less topography is flat in all sites. than 500 m; 2. Semi-isolated (other trees no The Lagartera trial is found on fine clay closer than 100 m); and 3. Mother tree in clusters isohyperthermic soils of the Aquic Distropept or associated with more than 2 trees in a radius of family. These soils, characteristic of old terraces less than 100 m. This variable was analyzed with with little evolution, tend to be slightly acid, and orthogonal contrasts, analysis of variance and the high clay content may cause drainage Tukey means comparisons. problems. We examined a total of 92 families from 7 The Upala trial is located on fine clay populations: Mexico, Guatemala, Belize, isohyperthermic, reddish soils of the Oxic Honduras, Nicaragua, Costa Rica and Panama. Dystropept family. Such soils are deep, but The 3 field trials are all in the life zone low in bases and associated with poor referred to by Holdridge (1967) as Tropical drainage. NAVARRO Y HERNÁNDEZ: Population differentiation of Mahogany 43

Table 4. Families planted in trial at Laberinto, Los Chiles, Costa Rica.

Country Population Number of Family Latitude Longitude Precipitation Temperature Dry GPS GPS (mm) (°C) months

MEXICO San Felipe 11, 15, 16, 116 18.74599 88.3548 1300 25 4 MEXICO Nuevo Becal I 121, 122, 125, 126, 129 18.80756 89.32721 1200 26 4 MEXICO Nuevo Becal II 156, 157,159,160 18.80756 89.32721 1200 24 4 MEXICO Naranjal 132, 133, 136, 141 19.35549 88.46355 1200 24 4 MEXICO Madrazo 152, 153, 155 19.45218 88.44468 2000 26 4 BELICE Las Cuevas I 22 16.7518 89.00106 2900 22 3 BELICE Las Cuevas II 257 16.42754 88.58995 2900 22 3 GUATEMALA Bethel 32, 326, 331, 343 16.4835 90.50282 1800 25 4 GUATEMALA Bio-Itza 396 16.85234 90.93113 1955 28 5 GUATEMALA Tikal I 3101 16.85009 90.9316 1955 28 5 GUATEMALA Tikal II 3131, 3151, 3153 17.22520 89.61216 1955 28 5 HONDURAS Lancetilla 427, 432, 434, 435, 436, 438, 15.73991 85.45721 3278 25 3 440, 442, 443, 444, 445, 447 HONDURAS Comayagua I 453 14.45397 87.6597 1619 25 5 HONDURAS Comayagua II 454 14.45398 87.65806 1619 25 5 NICARAGUA Mukuwas 527, 528, 529, 531, 532, 533, 14.04645 84.49976 2750 24 4 536, 541, 551, 556, 559, 560, 561, 562, 564, COSTA RICA Caño Negro 644 *, 649 * 10.97774 84.70994 2885 24 3 COSTA RICA Marabamba 682 * 10.95403 84.65751 2885 24 3 COSTA RICA San Emilio I 687* 10.97235 84.77287 2885 24 3 COSTA RICA San Emilio II 690* 10.95402 84.77411 2885 24 3 COSTA RICA San Emilio III 693* 10.95971 84.77346 2885 24 3 COSTA RICA Upala I 699 10.54085 85.09293 2558 25 4 COSTA RICA Santa Cecilia 6109 11.06101 85.27857 2585 26 4 COSTA RICA Abangares 6121 10.05493 84.49443 1940 27 5 COSTA RICA Pocosol 6156 10.89688 85.60125 1510 26 6 COSTA RICA Chapernal 6243, 6244, 6248, 6250, 6251 10.06589 84.53552 1940 27 5 COSTA RICA Upala II 6253 10.84879 84.92218 2558 25 4 PANAMÁ Quintin 73, 79, 710 8.25682 78.26795 2500 26 4 PANAMÁ Punta Alegre I 711 8.04698 78.23505 2500 26 4 PANAMÁ Punta Alegre II 713 8.26119 78.23616 2500 26 4 PANAMÁ Tonosi 717, 719, 721, 724, 726 7.33517 80.48316 2500 25 4 PANAMÁ Gatún 731, 732, 733 9.268 79.91958 2500 25 4 PANAMÁ Paraíso 735 9.03278 79.62656 2500 25 4 PANAMÁ Coiba 741 7.50102 81.69603 3500 25 4

* Also planted in trials at Upala and Lagartera.

Experimental design plots, details like number of families per block, families by site and blocks by site are described A randomized block design was applied in table 2. in all trials. Two tree family plots were planted Preparation of the area involved at 3x3 m with 2 guard rows around the whole plowing and raking the soil previous trial. In Upala trees were interplanted in lines plantation. The plots were maintained with with Calophyllum brasiliense at a 3x3 m machete use and the herbicide Roundup spacing. Only Lagartera have 3 trees per family (glyphosate). 44 AGRONOMÍA COSTARRICENSE

The models used for the statistical analysis BPij is the interaction of ith block with the family j. were: Blocks and families are considered as random effects. ● Random analysis of variance on individual trees Thus, one may expect different family µ Yijk = + Bi + Pj + BPij+ εjkl structures of the seeds from open-pollinated Yijk is the phenotypic value of the kth tree of the jth collections, sometimes self-pollinated due to family in the ith block; isolation. Therefore, families of single mother- µ = population mean; trees may be half-sibs, full-sibs or selfings.

Pj = effect of the jth family; The procedure (proc) in SAS for Bi = effect of ith block evaluating the last 2 measurements was: proc glm; class evaluation block population family; model diameter height attack respatack form = eval block population family(population) population*block eval*population*family family*eval(population); means population; means population/Tukey lines; where eval is evaluation number and respattack is response to H. grandella attack.

The proc for the evaluation of the mother tree isolation for both the statement Contrast of GLM and Tukey means comparison was the following: proc glm; class block mothertree; model diameter height attack shoots form =block mothertree; means mothertree/Tukey lines; contrast ‘1 vs 2y3’ mothertree 2 -1 -1; contrast ‘2 vs 3’ mothertree 0 -1 -1;

The following parameters for growth in Coefficient of population differentiation was height and diameter were estimated based on calculated according to Kremer et al. (1997). individual mother trees: the additive genetic σ2 σ2 σ2 variance A = 4 F, where F is the variance component due to open-pollinated families; the RESULTS AND DISCUSSION narrow-sense heritability on an individual tree 2 σ2 σ2 σ2 σ2 σ2 basis h = A / ( F + BXF + E), where E is Highly significant differences between the residual variance. The standard errors for families and populations for the variables height heritability were calculated following the and diameter were found in the Laberinto trial formula by Dieters et al. (1995). The additive (P<0.0001), (Tables 5 and 6). Differentiation of genetic coefficient of variation AGCV=100 the populations is shown in the cluster of figure σ ( A/x), where x is the population mean. 2, a clear bifurcation between populations of NAVARRO Y HERNÁNDEZ: Population differentiation of Mahogany 45

Table 5. Variance components and significancies at Laberinto trial (Los Chiles, Costa Rica).

Variables Effect Block Pop Fam BlockPop Error Among Within Qst (Pop) pops pops

Diameter 1 VC 7.636 2.245 3.756 1.402 36.376 37.41 62.58 0.0695 p <.0001 <.0001 <.0001 <.0001 Diameter 3 VC 27.134 24.623 7.446 12.1144 222.282 76.78 23.21 0.2924 p <.0001 <.0001 0.0062 0.0036 Diameter 4 VC 22.985 42.898 5.523 6.177 221.83 88.59 11.40 0.4926 p <.0001 <.0001 0.0219 0.0329 Height 1 VC 177.85 21.751 143.882 7.53 975.659 13.13 86.86 0.0185 p <.0001 <.0001 <.0001 0.0439 Height 2 VC 581.821 741.867 208.146 358.877 4773.9 78.09 21.90 0.3082 p <.0001 <.0001 0.0002 <0.0001 Height 4 VC 1231.8 3166.1 321.182 686.423 9888.4 90.78 9.21 0.5520 p <.0001 <.0001 0.0121 0.001 H.g. resistance 1 VC 0.000187 0 5.48E-06 0.000371 0.0124 0 100 0 p <.0001 0.0711 0.6866 0.0001 H.g. resistance 4 VC 0.000203 7.33E-05 0.000236 0.000323 0.0094 23.70 76.29 0.0373 p <.0001 <.0001 0.0124 0.0008 Stem form 2 VC 0.007691 0.00314 0 0.0018 0.287 100 0 1 p 0.0021 0.1529 0.2995 0.1033 Stem form 4 VC 0.0279 0.04112 0.01147 0.05461 0.6193 78.18 21.81 0.3094 p <.0001 <.0001 0.0065 <.0001 # shoots 2 VC 0.02278 0.01472 0.04602 0.02019 0.6895 24.23 75.76 0.0384 p 0.0131 <.0001 <.0001 0.0683 # shoots 4 VC 0.0686 0 0.0145 0.0026 0.642 0 100 0 p <.0001 0.302 0.0326 0.076

Number after variables indicates dates of measurement: 1. 251 2. 979 3. 1054 4. 1327 days after planting

Table 6. Results of ANOVA, expected mean squares and genetic parameters for families of S. macrophylla, at Laberinto, Costa Rica.

Mean Genetic Mean Genetic Mean Genetic (CV %) parameters (CV %) parameters (CV %) parameters

1998 1998 2000 2000 2001 2001 Diameter 19.8 (28) h2=0.55, AGCV=13 63 (24) h2=0.35± 0.01 71 (22) h2=0.48±0.01 Height 97.5 (30) h2=0.57, AGCV=14 267 (26) h2=0.41± 0.007 328 (31) h2=0.6±0.01 Attack 0.015 (720) h2=0.02 1 (0) NC 0.98 (9) h2=0.1±0.002 # Shoots after 2.18 (38) h2=0.27± 0.004 2.5 (31) h2=0.07±0.002 attack of H. grandella Stem Form 1.12 (47) h2=0.04± 0.001 1.4 (57) h2=0.18±0.003

Costa Rica and Panama and the rest of components indicate differences among and Mesoamerica shows the genetic structuring of within populations. At 251 days measurement the species, these results are comparable with the variation within populations was greater than ones obtained using molecular markers by variation among populations, but for Gillies et al. (1999). The attack by H. grandella measurements at 979, 1054 and 1327 days, that showed no significant differences among condition changed, showing more variation families in the Laberinto trial. In table 5 variance among populations than within populations. 46 AGRONOMÍA COSTARRICENSE

Santa Cecilia among populations has very important Naranjal implications for gene conservation and breeding, Marabamba as it suggests that local breeding activities would Bio-Itza have high possibilities of securing substantial San Felipe genetic gains. Mukuwas Table 6 shows the heritabilities obtained Lancetilla for diameter and height at the Laberinto trial San Emilio (0.55 and 0.57, respectively), and the additive Madrazo genetic coefficient of variation (AGCV) with Tikal 13% for diameter and 14% for height. Pruning Nuevo Becal was applied to the experiment immediately after Las Cuevas the measurement for the year 2000. Our Bethel heritability estimates are high in comparison Punta Alegre population with the ones shown by Cornelius (1994), who Pocosol obtained mean heritabilities of 0.28 for height Gatún and 0.23 for diameter in a total of 67 studies, Comayagua Coiba mainly with pines. For AGCV, Cornelius (1994) Quintin obtained a range of 5 to 15%; therefore, the Tonosi present study suggests high levels of additive Paraíso genetic variation for height and diameter. Caño Negro The heritability in the production of shoots Upala after attack by H. grandella is larger in year Chapernal 2000, mainly because before that year trials were Abangares maintained without sanitary pruning. After the measurement of the year 2000, trees attacked by 2.5 2.0 1.5 1.0 0.5 0.0 H. grandella were pruned, reducing the number McQuitty’s Similarty of shoots. In the measurement at 251 days at Fig. 2. Cluster for Swietenia macrophylla populations in Laberinto, families 721 and 444 from Panama Mesoamerica. Laberinto site (Los Chiles, Costa and Honduras, respectively, presented the best Rica) at 3.65 years. diameter increment (27 mm). Families 444 and 564 (Nicaragua) excelled in height (130 cm); these results were obtained using the Variation between different years of measurements at 118 days as co-variables to measurement was found for the coefficient of eliminate the nursery or maternal effects in the population differentiation (Qst) at 251 days and ANOVA shown in table 7. the other measurements. Also maternal effects The best (721) and the worst (528) possibly are affecting Qst values for the families differed by 127% in collar diameter. For measurement at 251 days, where there is no height, the range from best (444) to worst (726) differentiation for populations. From the results was 131%. and comparison of means, it can be seen that The Gatun (Panamá), Comayagua populations show a variation between the best (Honduras) and Coiba (Panamá) populations and the lowest population of approximately 30% showed the lowest performance, being inferior and 21%, in diameter and height, respectively. At by 30% in diameter and 52% in height, compared the family level, the variation for diameter was to the best performers Bio-Itza and Santa Cecilia. 125% between the highest and lowest family, It can be observed that variation among while for height, the variation was 130%. The populations were lower than among families fact that there is more variation within than within locations. The trees from Bio-Itza NAVARRO Y HERNÁNDEZ: Population differentiation of Mahogany 47

Table 7. Covariance analysis for families of S. macrophylla, at 339 days. Laberinto, Costa Rica. Co-variables diameter and height at 118 days after the sowing of the seeds were introduced in the model.

Trait/Effect Degrees of freedom Mean Square F P >F Mean (CV %)

Diameter/Blocks (B) 14 302.29 12.35 0.0001 20.0 (21.3) Families (P) 91 64.109 2.62 0.0001 B * P 1229 24.469 1.35 0.0001 D118 1 16853.27 688.7 0.0001 Error 1214 18.1885 height/Blocks (B) 14 7502.00 12.70 0.0001 98.5 (21.7) Families 91 1420.362 2.40 0.0001 BxP 1229 590.7621 1.29 0.0001 Alt 118 1 453304.9 767.3 0.0001 Error (E) 1214 456.599

(Guatemala), Tikal (Guatemala), Santa Cecilia more than 100% superiority in height (Costa Rica), Marabamba (Costa Rica) and increment to Calophyllum brasiliense. This Naranjal (México) came from natural stands reflects the different ecological behavior of the composed by large populations; this is reflected two species, mahogany being a light tolerant in the superior performance of their families (381 pioneer while Calophyllum is a shade tolerant cm of height and 77 mm of diameter in average climax species. for the five populations at 3.6 years). The individual narrow sense heritability Table 8 shows the analysis of variance estimates, for diameter and height, were 0.54 and for the family trial in Upala. As in the other 0.55, respectively in Upala trial. trials, significant differences were found The family with the best growth was No. 93 (P>0.0001) in diameter and height but not for from San Emilio (see Table 3 for details), an the variable Hypsipyla attack. The growth of important area of natural forest in northern Costa mahogany is overwhelming in this location Rica. This family was superior by 45% to family (growth can be considered fast when root collar 76 from Marabamba. Family 93 was 50% superior diameter increases by 2 cm per year and height in diameter and height, compared to family 45 of more than 1 m per year). Mahogany registered Caño Negro (Costa Rica). These data show that

Table 8. Expected square means and genetic parameters for the trial of Swietenia macrophylla in Upala, Costa Rica at 621 days.

Trait Effect Mean square P >F Variance Mean Genetic component (CV %) parameters

Upala Blocks (B) 1176.3 0.0001 20.85 55(14.6) h2=0.54 AGCV=7.4 Diameter Families (P) 529.2 0.0001 17.45 AGCV=8 138.34 0.0001 40.15 65.53 66.38 Upala Blocks (B) 29752.5 0.0001 567.68 246 h2=0.55 AGCV=7.8 Height Families(P) 12056 0.0001 373.39 (15) AGCV=8 BxP 2987.3 0.0001 850.82 Error (E) 1466 1466.88 Upala Blocks 0.61935 0.0001 0.008 0.8 h2=0.065 Attack Families 0.214551 0.0002 0.002 (36) BxP 0.149037 0.0001 0.027 Error 0.094623 0.094 48 AGRONOMÍA COSTARRICENSE local material of northern Costa Rica contain Table 10 presents the analysis of variance, enough variation for an improvement program. table 11 the orthogonal contrast analysis, and the Table 9 shows the results of the Lagartera test of Tukey for the means. We interpret this fact trial where flooding affected some blocks in the as a possible indication of an increase in selfing test. The effect of families was not significant for in case of solitaries. Karkkainen et al. (1996) and diameter but significant for the variables height Koski and Muona (1986) in studies with Pinus and Hypsipyla attack (P>0.01). sylvestris have showed the same possible effect Lagartera at 1812 days (4.6 years) had a of inbreeding. Future tree improvement diameter average of 68 mm and a coefficient of programs must consider the risk of inbreeding variation (CV) of 21%. Diameter differences depression when collecting seed from single were significant to Pr > F =0.0017. trees growing far apart. The mating system, The family 50 had the best diameter (78.8 including estimates of selfing, should be mm), and the lowest performance was for family undertaken using relevant genes markers. 68 with a diameter of 49.2 mm. The variable We suggest that family differences reflect height registered an average of 306 cm, while the 3 sources of variation: 1) among populations; 2) CV was 31% (Pr > F =0.0262). Again, the best among families within populations due to family was the 50 with 306.8 cm and the lowest additive genetic variability; and 3) among family the 68 with 188.8 cm, at 4.5 years. families due to inbreeding depression. Pruning was applied 1 year before the 2001 These differences clearly reflect measurement. adaptations to different environments such as the Analysis of variance is presented in tables 10 latitude spans from 7.5 degrees to 19.45 degrees, and 11. Results show that solitary (isolated) mother precipitation from 1200 to 3500 mm year-1 and the trees presented lower progeny performance in length of the dry season from 3-6 months per year. height and diameter (Pr > F <0.0001 and form The variation in the Fst values in the Pr > F <0.0037). measurement at young stages may be caused by The presence of attack was not maternal effects; the same applies for the significant in the performance of the family for components of variance within and between the solitary analysis; nor was the response to the populations. Also, for the first measurement, all attack (Pr > F =0.3440). It is interesting to note trees had the same soil condition in the plastic pot, that there appears to be differences due to the so the adaptive characters given to different population structures. Thus, mother trees that microenvironments were minimized. The other grow together in clumps generally show superior measurements were done in the field where the families when compared to the families from seedlings were submitted to the edaphic and climatic mothers that grow as solitaries. conditions in the northern zone of the country.

Table 9. Results of ANOVA, expected square means and genetic parameters for the trial Swietenia macrophylla La Lagartera, Costa Rica at 585(1998) and 4.54 (for height) and 4.96 (for diameter) years after planting.

Mean Genetic Significance Mean Genetic Significance (CV %) parameters (CV %) parameters

1998 1998 2001 2001 Diameter 47 h2=0.19±0.009 0.0011 68 h2=0.46±0.03 <.0001 (24) AGCV=11 (21) Height 182 h2=0.24± 0.01 0.0001 306 h2=0.35±0.02 <.0001 (27) AGCV=14 (31) Attack 0.68 h2=0.14± 0.001 0.0011 0.2 h2=0.13±0.01 0.0111 (65) AGCV=26 (475) NAVARRO Y HERNÁNDEZ: Population differentiation of Mahogany 49

Table 10. Analysis of variance for the variable solitary mother tree at 3.6 years. Laberinto trial. Los Chiles. Costa Rica.

Source DF Type III SS Mean Square F Value Pr > F

Diameter Block 14 76411.11960 5457.93711 19.48 <.0001 Mother tree 2 26488.79272 13244.39636 47.26 <.0001 Error 3386 948820.150 280.219 Height Block 14 2772736.400 198052.600 20.11 <.0001 Mother tree 2 1312080.358 656040.179 66.60 <.0001 Error 3392 33410485.04 9849.79 Attack Block 14 0.17365414 0.01240387 2.52 0.0014 Mother tree 2 0.04247855 0.02123927 4.31 0.0134 Error 3392 16.69869642 0.00492296 Form Block 14 45.39012108 3.24215151 6.28 <.0001 Mother tree 2 5.78438930 2.89219465 5.61 0.0037 Error 3392 1750.014126 0.515924 # of shoots Block 14 134.3258932 9.5947067 12.93 <.0001 Mother tree 2 2.2876912 1.1438456 1.54 0.2142 Error 3393 2517.478661 0.741962

Table 11. Orthogonal contrasts for the variable solitary mother tree and Tukey grouping for the analysis of the variable solitary mother tree. Laberinto trial. Costa Rica.

Contrast Analysis Mean Square F Value Pr > F Tukey Groups Mean (cm) Mother tree*

Diameter Diameter Contrast A 70.7063 2 1 vs 2 y 3 24276.68368 86.63 <.0001 A 68.9839 3 2 vs 3 831.70919 2.97 0.0850 B 63.7809 1 Height Height 1 vs 2 y 3 1123677.466 114.08 <.0001 A 317.624 2 2 vs 3 9069.200 0.92 0.3373 A 312.806 3 B 274.469 1 Attack Ataque 1 vs 2 y 3 0.02972101 6.04 0.0141 A 0.998144 3 2 vs 3 0.00049464 0.10 0.7513 A 0.997354 2 A 0.990813 1 # of shoots # of shoots 1 vs 2 y 3 0.87230015 1.18 0.2783 A 2.37359 1 2 vs 3 0.40331726 0.54 0.4610 A 2.35450 2 A 2.31374 3 Form Form 1 vs 2 y 3 3.92386213 7.61 0.0059 A 1.31802 1 2 vs 3 0.09892198 0.19 0.6615 B A 1.24339 2 B 1.23082 3

* 1. Isolated mother tree (no other trees of the same species) at a distance of less than 500 m, 2. Semi-isolated (other trees no closer than 100 m) and 3. Mother tree in clusters or associated with more than 2 trees in a radius of less than 100 m. 50 AGRONOMÍA COSTARRICENSE

Gillies et al. (1999) in a Mesoamerican Rica) and 51 from Caño Negro (Los Chiles, study of mahogany using RAPD, found a Costa Rica) produced the best growth in height significant portion of diversity maintained and diameter. between populations within the three geographical As for resistance to H. grandella, no regions in Mesoamerica (i.e. Yucatán Península, significant variation was found in the presence of Central Zone and Panama); the present results also the shootborer in any of the 3 trials. These results show that there is differentiation in clusters of indicate that breeding for resistance in the Panama in comparison with the other present material is probably worthless. We populations in the northern part of Mesoamerica believe, the best strategy to avoid insect damage (Figure 2). The percentage of genetic variation is integrated pest management through the use of among populations was 12%, Fst 0.0168, while mixed stands, and agroforestry including on average for 12 values of different characters important crop plants. and measurements of Fst values was 0.2. These The heritability estimates were relatively differences show the influence of the selection in high for diameter and height, which suggest the Swietenia macrophylla because of great climatic possibility of good genetic gains and indicate that and edaphic differences within the entire there is additive genetic variability available for collection. successful breeding. The criteria used to cut the adult trees in The collection of seeds from individual the natural forest are: a minimum of diameter; trees to grow families has given us a reason to height; good form; and the condition of trees suggest further careful studies on population without a hollow trunk. In the analysis we structures and especially on risks of inbreeding in ascertained that the trees in exploited areas have sparse populations that may be the result of long- less performance in height, diameter and form, term forest exploitation. We recommend and that hollow trunks is a variable that can not combining quantitative studies with molecular be analyzed in young trees. Gillies et al. (1999) markers, including mating systems, in order to found that the genetic diversity in mahogany was save valuable genetic resources. Only after such correlated to the level of exploitation or information it will be possible to plan optimal destruction of the forest; these results can be tree breeding programs for mahogany. related with those presented in this paper, given that the solitary trees presented a low performance in diameter and height. This could ACKNOWLEDGEMENTS indicate that the dysgenic selection produced by the human exploitation of the best trees, could To the European Union, contract TS3*- have reduced the variation that corresponds to CT94-0316 and of the IC18-CT97-0149 for its that part of the populations, including the best support. M. Hernández, J. Cornelius, L. Coto, J. model trees, and therefore decreasing the overall Perez, K. Wightman and J. Haggar, all from genetic diversity. CATIE, J. Wilson and A. Gillies from the Center Progeny experiments made with both for Ecology and Hydrology (Scotland) and the Costa Rican families (at Upala and Lagartera), landowners R. Nuñez, J. E. Rodríguez and H. and the Central American and Mexican trial (at Abarca are gratefully acknowledged. In Central Laberinto), reveal high levels of genetic variation America we thanks to Centro Maya, Parque for height and diameter growth on 2 levels, Nacional Tikal, ESNACIFOR, Centro de population and family. For diameter, the best Mejoramiento Genético and Semillas Forestales, populations were Bio-Itza (Guatemala) and Tikal INRENARE, Proyecto CATIE-CONAP, Forest (Guatemala) and for height, Santa Cecilia (Costa Dept. of Belize, Las Cuevas Experimental Station, Rica) and Naranjal (México). and in Mexico to ICRAF and INIFAP for their For the Upala trial, it was found that cooperation. To Professor Peter Tigerstedt for his families 93 from San Emilio (Los Chiles, Costa valuable comments and suggestions for this paper. NAVARRO Y HERNÁNDEZ: Population differentiation of Mahogany 51

LITERATURE CITED embryonic lethals. pp.391-400. In: IUFRO. Conference Proceedings Breeding Theory. Oct. 13- BOONE R.S., CHUDNOFF M. 1970. Variations in wood 17. 1986. Williamsburg, Virginia. density of the mahoganies of Mexico and Central America. Turrialba 20 (3):369-371. KREMER A., ZANETTO A., DUCOUSSO A. 1997. Multilocus and multitrait measures of differentiation CITES. 1997. Proposed amendment to CITES Appendix II. for gene markers and phenotypic traits. Genetics Convention on international trade in endangered 145: 1229-1241. species of wild fauna and flora. 10th Meeting of the Conference of the Parties, Harare, Zimbabwe, 9-20 MATAMOROS Y., SEAL U.S. (eds.). 1996. Report of June 1997. 36 p. threatened plants of Costa Rica Workshop, 4-6 October. IUCN/SSC Conservation Breeding CORNELIUS J.P. 1994. Heritabilities and additive genetic Specialist Group: Apple Valley MN. coefficients of variation in forest trees. Can. J. For. Res. 24(2):372-379. NAVARRO C., WILSON J., GILLIES A., HERNÁNDEZ M. 2003. A new Mesoamerican collection of big-leaf DIETERS M.J., WHITE T.L., LITTELL R.C., HODGE G.R. Mahogany, pp 103-117. In: A.E. Lugo, J.C. Figueroa 1995. Application of approximate variances of and M. Alayón (eds.). Big leaf mahogany: Genetics, variances components and their ratios in genetic ecology and management. Ecological Studies 159, tests. Theor. Appl. Gen. 91:15-24. Springer. New York.

GILLIES A., NAVARRO C., LOWE A.J, NEWTON A.C., NEGREROS C., MIZE C. 1994. El efecto de la abertura del HERNANDEZ M., WILSON J., CORNELIUS J.P. dosel y eliminación del sotobosque sobre la 1999. Genetic diversity in Mesoamerican regeneración natural de una selva de Quintana Roo. populations of mahogany (Swietenia macrophylla), pp. 107-126 In: Snook, L. & A. Barrera de Jorgenson assessed using RAPDs. Heredity 83 (1999): 722- (eds.). Madera, Chicle, Caza y Milpa: Contribuciones 732. al manejo integral de las selvas de Quintana Roo. INIFAP/PROAFT/AID/WWF-US, Mérida, México. GRAHAM A. 1999. Studies in Neotropical Paleobotany. XIII. An Oligo-Miocene Palynoflora from Simojovel PATIÑO F. 1997. Genetic resources of Swietenia and Cedrela (Chiapas, Mexico). American Journal of Botany in the neotropics: Proposals for coordinated action. 86(1): 17-31. 1997. Based on contractual work for FAO by P.Y. Kageyama, C.Linares, C. Navarro and F. Patiño. GULLISON R. E., PANFIL S.N., STROUSE J.J., HUBBELL Forest Resources Division, Forestry Department, S.P. 1996. Ecology and management of mahogany FAO, Rome. (Swietenia macrophylla King) in the Chimanes Forest, Beni, Bolivia. Botanical Journal of the PÉREZ S., ALVARADO A., RAMÍREZ C. 1978. Mapa Linnean Society 122:9-34. asociación de subgrupo de suelos de Costa Rica. San José, Costa Rica, Instituto Geográfico Nacional. Esc. HOLDRIDGE L.R. 1967. Life zone ecology. San Jose, Costa 1: 200.000. Rica. Tropical Science Center. 206 p. SNOOK L.K. 2003. Natural regeneration and growth of KARKKAINEN K., KOSKI V., SAVOLAINEN O. 1996. Mahogany (Swietenia macrophylla) in Mexico´s Geographical variation in the inbreeding depression Yucatan forests: Implications for sustainable of Scots Pine. Evolution 50(1): 111-119. silviculture, pp.169-192. In: A.E. Lugo, J.C. Figueroa and M. Alayón (eds.). Big leaf Mahogany: KOSKI V., MUONA O. 1986. Probability of inbreeding in Genetics, ecology and management. Ecological relation to clonal differences in male flowering and Studies 159, Springer. New York.